Thermal Rearrangement of Allyl Vinyl Ether - American Chemical

Thermal Rearrangement of Allyl Vinyl Ether: Heavy-Atom. Kinetic Isotope Effects and the Transition Structure. Lidia Kupczyk-Subotkowska,l William H. S...
0 downloads 0 Views 573KB Size
5957

J. Am. Chem. SOC.1993,115, 5957-5961

Thermal Rearrangement of Allyl Vinyl Ether: Heavy-Atom Kinetic Isotope Effects and the Transition Structure Lidia Kupczyk-Subotkowska,l William H. Saunders, Jr.,*J Henry J. Shine,’J and Witold Subotkowski’ Contributionfrom Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, Texas 79409, and Department of Chemistry, University of Rochester, River Station, Rochester, New York 14627 Received December 9, 1992

Abstract: Kinetic isotope effects (KIE) in the rearrangement of allyl vinyl ether (1,3-oxa-1,5-hexadiene) to 4-pentenal (2) were measured for labeling with 14C at the 2-, 4-, and 6-positions and with 1 8 0 at the 3-position. BEBOVIB modeling calculations were applied successfully to the heavy-atom KIE and previously reported KIE for deuterium substitution at positions 4 and 6. From the calculations, it is deduced that the C4-0 bond is 50-70% broken and the C1-Q bond 10-30% formed in the transition structure. Further, there is strong coupling between the allyl and vinyloxy fragments, strong bonding within the allyl but relatively weak bonding within the vinyloxy fragment. Introduction The thermal, aliphatic Claisen rearrangement has evoked considerableinterest in its applicationsto syntheticand theoretical organic chemistry.’ The prototype rearrangement of allyl vinyl ether (1) to 4-pentenal (2), particularly, has been the focus of theoretical and experimental works that have sought to characterize its transition structure. The rearrangement is formally a [3,3]-sigmatropicshift. The theoretical interest in it has been centered on the nature of the transition state and on whether rearrangement is concerted or is a two-step one in which a 1,4diradical intermediate participates. Coates and co-workers4 measured the rates of rearrangement of 1 and 4-, 5-, and 6-methoxy-1 in several solvents and concluded from the effects of solvent and substituentsthat the rearrangementof 1is concerted and has a pronounceddipolar character. Insofar as concertedness is concerned, there was, initially,somecontroversy over the timing of the bond processes. Based on measurement of deuterium kinetic isotope effects (KIE) in rearrangements of [4-2H~]-1 and [6JH2]1, Gajewski and Conrad concluded that bond breaking was advanced over bond formati~n.~ This conclusion was supported with ab initio calculations by Houk and collaborators6 but contradicted by earlier MNDO calculationsof Dewar and Healy.’ Later, however, Dewar and Jie? with the use of AM 1calculations, concluded that the rearrangement does seem to follow the course suggested by Gajewski and Conrad. Recently, we reported carbon and oxygen KIE for the aromatic Claisen rearrangement of allyl phenyl ether.gJO Bond-order calculations were made to fit not only these six heavy-atom KIE but also the earlier deuterium KIE of McMichael and Korver.11 We were able to show that coupled motion within the six-atom pericyclic array led to KIE at the phenyl’s C1 and the allyl’s &carbon atom, that bond breaking (50-608) was advanced over (1) Texas Tech University. Dr. Subotkowski’s present address: Chemsyn Science Labs, Lenexa, KS 66215-1297. (2) University of Rochester. (3) Ziegler, F. E. Chem. Rw. 1988,88, 1423. (4) Coat=, R. M.; Rogers, B. D.; Hobbs, S. J.; Peck, D. R.; Curran, D. P. J. Am. Chem. Soc. 1987,109, 1160. (5) Gajewski, J. J.; Conrad, N. D. J. Am. Chem. Soc. 1979, 101,6693. (6) Vance, R. L.; Rondon, N. G.;Houk, K. N.; Jensen, H. F.; Borden, W. T.; Komornicki, A.; Wimmer, E. J . Am. Chem. Soc. 1988,110, 2314. (7) Dewar, M. J. S.;Healy, E. F. J. Am. Chem. SOC.1984, 106, 7127. (8) Dewar, M. J. S.; Jie, C. J. Am. Chem. Soc. 1989, 111, 511. (9) Kupczyk-Subotkowska, L.; Saunders, W. H., Jr.; Shine, H. J. J. Am.

Chem. Soc. 1988, 110,7153. (10) Kupczyk-Subotkowska, L.; Subotkowski, W.; Saunders, W. H., Jr.; Shine, H. J. J . Am. Chem. Soc. 1992,114,3441. (1 1) McMichael, K. D.; Korver, J. L. J. Am.Chem. Soc. 1979,101,2746.

Scheme 1”

aH

Hi4-

l 4 m

bH14COOC8H17

Reagents: (a) 85% HsP04and thendryingwith B2O3; (b) C S H ~ ~ O H , TsOH; (c) CpZTiMez.

bond forming (10-20%), and that transition-state coupling was stronger between the three-atom fragments rather than within them. We have now applied this methodology to the rearrangement of 1 (eq 1). We have measured KIE for rearrangement of [23 0

4

1

5 (1)

(2)

14C]-,[4-W]-, [6-14C]-,and [1*0]-1and havemadecut-offmodel calculations to attempt to fit these and Gajewski’s deuterium5 KIE to a model of the transition structure.

Experimental Section Preparationof AUyl Vhyl Ether (1). n-Octylvinyl ether was prepared from commercial ethyl vinyl ether and n-octanol using the procedure of Watanabe and Conlon12 and then used as the vehicle for preparing 1, as follows. A mixture containing 12.9g (82.6 mmol) of n-octyl vinyl ether, 4.0 g (68.9 mmol) of allyl alcohol, and 0.5 g of mercuric acetate was heated so that 1 distilled slowly through a 20-cm Vigreaux column and gave 4.2 g of crude 1, bp 58-70 OC. The distillate was washed with water three times to remove allyl alcohol and dried over KzC01. Bulb-to-bulb distillation under vacuum gave 1 which was redistilled under normal pressure to give 3.1 g (36.8 mmol, 54%) of 1, bp 64 OC. The product had satisfactory IH NMR and NMR spectra. Preparations of [4-14C]-1, [6-I4C]-1,and [180]-1were carried out, as described above, in reactions of n-octyl vinyl ether with [1-l4C]-, [314C]-, and [ 1-180]allylalcohol, whose syntheses, themselves, have been reported earlier.” [2-l4C]-1 was prepared similarly from reaction of allyl alcohol with n-octyl [2-W]vinyl ether. ~Octyl[2-14C]viylEther. This compound was prepared as shown in Scheme I. A mixture of 30 mg of sodium [14C]formate (American

(12) Watanabc, W. H.; Conlon, L. E. J . Am. Chem. Soc. 1957,79.2828. (13) Kupczyk-Subotkowska. L.; Shine, H. J. J. Labelled Compds. Radiopharm. -1*2, 31, 381.

OOO2-7863/93/1515-5957$04.00/00 1993 American Chemical Society

5958 J . Am. Chem. Soc., Vol. 115, No. 14, 1993 Radiolabeled ChemicalsInc., 5OmCi/mmol) and 2.69g of sodium formate (40.0 mmol) in 5.5 mL of 85% phosphoric acid was stirred overnight and then distilled bulb-to-bulb under vacuum to give 3.5 g of a mixture of [14C]formicacid and water. Part of the water was removed from this mixture by stirring overnight with 8.3 g of boric anhydride. Bulb-to-bulb vacuum distillation gave 2.21 g of product representing 1.84 g of formic acid, having a calculated activity of 50 mCi/mol, and 0.37 g of water. This product was esterified by refluxing with 10 mL (64 "01) of n-octyl alcohol and 5 mg of p-toluenesulfonic acid in 150 mL of benzene while a Deanatark trap was used to collect water. After reaction was complete (24 h), the benzene was evaporated and the product was purified by column chromatography on silica gel (J. T. Baker, 230-400 mesh) with diethyl ether/petroleum ether 1:l mixture as eluant. Distillation under normal pressure gave 3.89 g (24.6 mmol, 61%) of n-octyl [I4C]formate, bp 158 OC. After distillation, 1.O gof unlabeledn-octyl formate was placed in the distillation apparatus and distilled into the labeled product, giving 4.76 g (30.1 "01) of n-octyl [%]formate, having a calculated activity of 41 mCi/mol. The labeled n-octyl formate was converted into n-octyl [2J%]vinyl ether with the method described by Petasis and B z ~ w e for j ~ ~the methylenation of dodecyl acetate. To a solution of 7.2 g (34.6 mmol) of dimethyltitanoccnels in 150 mL of anhydrous tetrahydrofuran was added 2.36 g (14.9 mmol) of n-octyl [Wlformate, and the mixture was refluxed for 22 h while protected from light. The THF was concentrated by evaporation, and petroleum ether was added to precipitate a yellow solid. The solid was removed, and the filtrate was evaporated under reduced pressure. The residue was immediately purified by passing through a Florisil column with hexane as eluant to give 1.16 g (7.4 mmol, 50%) of n-octyl [2-W]vinyl ether. Unlabeled n-octyl vinyl ether (4.60 g) was added to the labeled compound, and the mixture was distilled under reduced pressure togive 5.47 g (35.0 mmol) of n-octyl [2-14C]vinyl ether, bp 115-1 16 OC, having a calculated activity of 8 mCi/mol. " h e r d Reamngement of Labeled 1. A standard procedure was used to carry out rearrangements. The treatment of the product depended on the isotopic label and the method of isotope assay. In the case of labeling, product 4-pentenal(2) was converted into its dimedonederivative. In the case of 1 8 0 labeling, the product was reduced to 4-pentenol which was then converted to its 1-naphthylurethane. Examples of procedures follow. Rmmgement of [ 4 - l Q I . (A) Law Conversion. A sample of 362 mg (4.30 mmol) of [4-1%]-1 was sealed under argon in a snap-neck ampule, which, before use, had been soaked overnight in KOH/ethanol mixture, washed with distilled water, and dried at 160 OC. The sealed ampule was placed in a thermostatically controlled oven at 160 OC. The ampule was removed after 60 min, cooled quickly, and opened. A 20-pL sample of the reaction mixture was used for 'H NMR spectroscopyfrom which the extent of rearrangement was calculated to be 30%. Assay of 1 was made by using the multiplet (dd) for CHz=CHO-allyl at 6.4 ppm and of 2 by using the triplet for -CHO at 9.8 ppm. All of the reaction mixture was next dissolved in 25 mL of ethanol and transferred into a round-bottomed flask which contained 470 mg (3.35 mmol) of dimedone, 2 drops of piperidine, and 1 mL of water. The flask was connected to a distilling head and heated. Unused 1 and 2-3 mL of ethanol distilled off. The solution remaining in the flask was cooled to room temperature, and 10 mL of water was added to precipitate a white solid which was filtered off. After the solid was air dried, 325mg (9.94 mmol, 73% based on 30% conversion) of the dimedone derivative of 4-pentenal was obtained. The product was purified with column chromatography on silica gel (J. T. Baker, 230-400 mesh) with a diethyl ether/hexane 1:l mixture as eluant. Finally, the product was sublimed twice, yielding 245 mg (0.71 mmol), mp 97-99 OC (lit.ls mp 98 "C), for scintillation counting. (B) 10O%Conversion. A sample of 149 mg (1.77 mmol) of [4-W]-1 was sealed under argon and heated at 160 OC for 24 h. 'H NMR spectroscopy showed that conversion into 4-pentenal was complete. The product was dissolved in 20 mL of ethanol, and 640 mg (4.56 "01) of dimedone, two drops of piperidine, and 1 mL of water were added. The mixture was refluxed for 10 min. Workup, as described above, gave 405 mg (1.17 "01) of dimedone derivative, mp 98 OC. In all, six lowconversion rearrangements of [4J4C]-1 were carried out in three sets of two ampules. For each set of two low-conversion ampules, one highconversion ampule was used. The data for all runs are listed in Table I. (14) Petasis, N. A,; Bzowej, E. 1. J . Am. Chem. Soc. 1990, 112, 6392. (15) Claur, K.; Bestian, H. Uebigs Ann. Chem. 1962,654, 8 . (16) Hurd, C. D.; Pollack, M.A. J. Am. Chem. Soc. 1938.60, 1905.

Kupczyk-Subotkowska et al. Table I. Kinetic Isotope Effects (KIE) for the Thermal Rearrangement of Allyl Vinyl Ether (1) to 4-Pentcnal (2) at 160 OC

run 1 2 3 4

isotope I

w

5

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

2-'4c

4-"C

6-I4C

conv 0.19 0.19 0.20 0.23 0.25 0.27 0.17 0.20 0.24 0.24 0.27 0.30 0.14 0.18 0.23 0.24 0.27 0.30 0.16 0.20 0.21 0.26 0.30 0.37

KIE 1.0503 i 0.0012 1.0507 i 0.0011 1.OS01 i 0.0020 1.0530 0.001 1 1.0483 i 0.0025 1.0515 i 0.0021 1.0264 i 0.0023 1.0240 i 0.0010 1.0289 i O.OOO9 1.0291 i 0.0014 1.0255 i 0.0018 1.0286 i 0.0006 1.0674 i 0.001 1 1.0704 i 0.0038 1.0758 0.0008 1.0780 i 0.0008 1.0776 i 0.0012 1.0627 i 0.0016 1.0196 0.0004 1.0165 i 0.0014 1.0156 i 0.0008 1.0184 i 0.0023 1.0191 i 0.0014 1.0176 i 0.0005

* *

Rearrangementof [lWo).l. (A) Low Conversion. Rearrangement was carried out as described for [4-I%]-1 with 535 mg (6.36 mmol) of allyl vinyl [18O]ether (12.9% enriched in l80) for 55 min at 160 OC. For the obvious reason, the dimedone method could not be used. The mixture containing 20% of [180]-2 and 80% of [180]-1 was dissolved in 15 mL of anhydrous ether and treated with 100mg of lithium aluminum hydride at 0 OC for 4 h. After reduction of 2 was complete (checked by TLC), 10 mL of 1 N sodium hydroxide was added and the ether layer which containtd4-[180]penten~landunreacted [180]-1 wasseparated. Separate experiments with unlabeled 1 showed that 1 was stable under these conditions. The aqueous layer was extracted twice with diethyl ether. The combined ether extracts was dried over K&O3 overnight. Ether and mostof [180]-1werecvaporatedunder reduced pressure toleave4-[180]pentenol. To this was added 2 15mg ( 1.27 mmol) of 1-naphthyl isocyanate, and themixturewasheated for 5 min. The l-naphthylurethaneof4-[~80]pentmol which precipitated was purified by column chromatography on silica gel (J. T. Baker, 60-200 mesh) with a diethyl ether/hexane 1:l mixture as eluant. The product was then crystallized from petroleum ether and sublimed twice to give 234 mg (0.92 mmol,72% based on 20% conversion), mp 55-58 OC, for mass spectrometry. (B) 100% Conversion. A sample of 1612 mg (1.91 mmol) of [180]-1 was sealed under argon and heated at 160 OC for 24 h to give 4-[180]pentenal. Reduction with 140 mg of lithium aluminum hydride gave 4-[180]pentenol, which was converted to its 1-naphthylurethane and purified as described above. KIE Measurements. Isotope abundanw for calculating oxygen KIE were obtained from whole moleculeion mass ratios measured on the 1-naphthylurethane of 4- [180]pentenol. A Hewlett-Packard quadrupole mass spectrometer, Model 5995, was used in the SIM mode. Samples were introduced via the solid sample inlet with a direct insertion probe. Relative abundanccs of the ions having m / z 255 and 257 were measured. Approximately 2000 scans were made per sample and analyzed in 25 blocks of 80 scans each. Carbon KIE were obtained from IC abundanccs measured with scintillation counting in a Bcckman Instruments counter, Model LS 5000 TD. A sample (about 9 mg) of the dimedone derivative of labeled 4-pentenal was weighed precisely (i0.002 mg) on a Cahn balance and dissolved in 10 mL of cocktail (Fisher Biotech ScintiLene BD, No. BP 455-4). Four samples (replicates) for each run were weighed. Thus, each measurement was an average of four samples. Each sample was counted 30 times (3 times per 1 cycle, 10 cycles). The 2a was sct at 0.5%.

Details of the methods and of calculating KIE have been described earlier." All KIE are listed in Table I, and the averaged KIE are given in Table 11. (17) Shine, H. J.; Kupczyk-Subotkowska, L.; Subotkowski, W.J. Am. Chem. Soc. 1985, 107, 6674.

J. Am. Chem. SOC.,Vol. 115, No. 14, 1993 5959

Thermal Rearrangement of Allyl Vinyl Ether Table II. Averaged KIE for the Rearrangement of 1 isdtope

KIE 1.0506 h 0.0007 1.0271 i 0.0006 1.0720 i 0.0008 1.0178 i 0.0005

1*0

2-14c 4% 6-l-

(4) absolute magnitudes, and the other was that couplings involvingthe same pairs of bonds in different models should be similar (they cannot be exactly the same between 3 and 4 on the one hand and 5 on the other because eqs 3 and 4 are different). We used the same criterion for goodnessof fit as before,1° namely, the quantity dev calculated from eq 5. This gives a quantity which behaves ((exptl

/NC

(5)

in the same way as a standard deviation, though its significance is, of course, different. Large deviationscontribute disporportionatelybecause of the squaring of the difference, which avoids a deceptively small value of dev in a case where the difference for one isotopic species is large but small for all of the others. The heavy-atom and H/Deffects were treated separately because the relative uncertainties in the latter are much larger than in the former. In order to give a sense of what the numbers in the dev columns of Tables 111and IV mean, the following artificial examples are provided. If all of the heavy-atom isotope effects were in error by a uniform f0.004, dev would be 0.0044, while an error of h0.003 would give dev = 0.0033. An error of h0.02in the secondary deuterium isotope effects would give dev = 0.0270, while an error of h0.03 would give dev = 0.0415.

3TS

3R

- calcd)/exptl)2/(n-1))1/2

Results and Discussion ITS

E

E ITS

5R

BEBOVIECalculrtiolr+ Themodelcalculationsusingthe BEBOVIB IV program in a version compiled to run on a Macintosh I1 computer's were carried out as b e f ~ r e . ~Models J~ 3R 315, 4R 415, and 5R Srs closely resembled those used for the aromatic Claisen rearrangement (Chart I). They follow the cutoff procedures (removal of atoms more than two bonds removed from thesiteof isotopicsubstitution) reported by Wolfsberg and Stern19*zo to give results close to those from full models. As before, we used different models for different patterns of isotopic substitution 80 as to avoid cyclic structures, which have been shown to introduce redundant coordinates that increase the difficulty of the calculations and complicate the interpretation of the results.21 Again,off-diagonal Fmatrix elementswereintroduced for the transition structures so as to ensure imaginary reaction coordinate frequencies in which bonding changes were occurring in the appropriate phases to transform the transition structures to products.22 The off-diagonal elementswere calculated from q 2. The relationship between am values for models 3 and 4 are given by q 3 and for model 5 by q 4. The a,,,,, values were varied 80 as to give the best fit to the experimental isotope effects for a given model, subject to two constraints. One was that 3TS, 4Ts, and SIS should have reaction coordinate frequencies of similar

-

-

~

+

~~

(18) Sims, L. B.; Burton, G. W.; Lewis, D. E. BEBOVIB-IV, QCPE No. 337;Department of Chemistry, Indiana University, Bloomington, IN 47405. (19) Wolfsbcrg, M.;Stem, M.J. Pure Appl. Chem. 1964, 8, 225. (20)Stem, M. J.; Wolfsbcrg, M.J. Chem. Phys. 1966,45,4105. (21)Keller, J. H.;Yankwich, P. J. Am. Chem. Soc. 1974,96, 2303. (22)Melander, L.; Saunders, W. H., Jr. Reaction Rates of Isotopic molecule^; Wiley-Interscience: New York, 1980; pp 64-66.

Experimental KIE. Substantial KIE were found for breaking the C-O bond (3-180and 4-14C) and for forming the C-C bond (6J4C). The results remove all question about the mechanism of this rearrangement. It is concerted, a two-step process in which one bond is made (or broken) before the other is broken (or formed) is ruled out. The large 3J80 and 4J4Ceffects (1.OS06 and 1.0720) and the smaller 6J4C (1.0178) effect suggest qualitativelythat cleavageof the carbon-oxygen bond contributes more strongly than the formation of the new carbon-carbon bond to the reaction coordinate motion. The experimentalisotope effect for the 2-C position (1.0271) demonstrates that, as with the aromatic Claisen case, experimental data can characterize the motion of atoms in a pericyclic array that are not themselves directly involved in bond breaking or forming. Thus, our data collectivelyshow that the six skeletal atoms of this rearrangement are coupled in motion. BEBOVIB Calculations. Our initial attempts were with five different sets of n(C0) and n(CC) values (the C-0 bond is breaking and the C-C bond forming in the transition structure) corresponding to those which gave the best fits for the aromatic Claisen rearrangement.1° The results in Table I11 are for a model that is closely similar in bonding to the best model for the aromatic Claisen rearrangement and uses the same sets of a,,,,, values. There are obvious and seriousdiscrepanciesbetween calculated and experimental values in Table 111. The calculated l80and 4J4C effects are much smaller and the calculated 4,4-D2and 6J4C effects are much larger than the experimentalvalues. After considerable trial and error, it became apparent that it was necessary to loosen the carbon-xygen bond in the developing carbonyl group and tighten the bonding within the allyl group relative to the aromatic Claisen model. Adjustment of the a,,,,, values was also required, in the direction of further increasing the strength of coupling between the two fragments (allyl and vinyloxy) in the transition structures, particularly at the 0 3 x 4 end. An especially troublesome problem was fitting the secondary deuterium isotope effects reported by Gajewski and Conrad.5 Initially, models that gave good fits to the heavy-atom isotope

5960 J. Am. Chem. Soc., Vol. 115, No. 14, 1993

Kupczyk-Subotkowska et al.

Table III. Calculated Isotope Effects at 160 OC for Model Rambling Aromatic Claisen' 0.510.2 0.5/0.l 0.410.3 0.310.3 0.410.2 exptl

1.0350 1.0350 1.0334 1.0318 1.0334 1.0506

1.1618 1.1618 1.1947 1.2292 1.1947 1.092

1.0354 1.0354 1.0373 1.0391 1.0373 1.0720

1.0338 1.0360 1.0353 1.0401 1.0386 1.0178

0.9339 0.9574 0.9360 0.9629 0.9602 0.976

1.0200 1.0178 1.0217 1.0214 1.0198 1.027 1

0.0243 0.0250 0.0239 0.0247 0.0248

0.075 1 0.063 1 0.0960 0.1 124 0.0875

0 omvalues 0.4,0.72,0.57, and 0.4 for model 3; 0.4,0.57, and 0.73 for model 4; and 0.4,0.72, and 0.83 for model 5. The order is that of the numbered bonds in the corresponding structures starting with CI and proceeding clockwise, skipping over bonds that are cut off in the models. Tunnel corrections calculated from the first term of the Bell equation are included in all isotope effects but change the effects by less than experimental error.

Table IV. Calculated Isotope Effects at 160 OC for Best Model' n(CO)ln(CC) 0.9/0.1 0.710.3 0.5/0.5

0.310.7 0.1/0.9 0.710.2 0.710.1 0.610.3 0.610.2 0.610.1 0.510.4 0.510.3 0.5/0.2 0.510.1 0.410.3 0.410.2 0.410.1 0.310.3 0.310.2 0.310.1 0.99/0.01b 0.01/0.99b exptl

180

1.0399 1.0398 1.0391 1.0372 1.0322 1.0475 1.0468 1.0474 1.0468 1.046 1 1.0470 1.0465 1.0459 1.0452 1.0453 1.0448 1.0442 1.0439 1.0434 1.0429 1.0004 1,0103 1.0506

4-Dz 1.0049 1.0645 1.1315 1.2082 1.2730 1.0643 1.0650 1.0779 1.0785 1.0792 1.1116 1.1122 1.1128 1.0945 1.1099 1.1105 1.1110 1.1278 1.1283 1.1287 1.0090 1.2874 1.092

4-1% 1.0489 1.0537 1.0580 1.0607 1.0575 1.0581 1.0613 1.0588 1.0620 1.0653 1.0609 1.0642 1.0674 1.0691 1.0659 1.0693 1.0727 1.0689 1.0723 1.0759 0.9999 1.0228 1.0720

6-'% 1.0268 1.0190 1.0155 1.0123 1.0096 1.0206 1.0225 1.0186 1.0202 1.0222 1.0167 1.0181 1.0197 1.0218 1.0175 1.0192 1.02'15 1.0169 1.0187 1.021 1 1.0015 0.9807 1.0178

6-Dz 1.079 1 0.9704 0.9317 0.8973 0.8668 0.9912 1.0136 0.9699 0.9907 1.0133 0.9500 0.9693 0.9902 1.0129 0.9687 0.9896 1.0125 0.9680 0.9890 1.0120 1.0046 0.7244 0.976

2-14c 1.0117 1.0135 1.0155 1.0181 1.0220 1.0274 1.0254 1.0286 1.0267 1.0246 1.0294 1.0277 1.0261 1.0240 1.0277 1.0258 1.0237 1.0277 1.0257 1.0237 0.9989 1.0009 1.0271

dev(hvy) 0.0173 0.0140 0.0120 0.0114 0.0141 0.0079 0.0068 0.0075

0.0060 0.0052 0.0065 0.0048 0.0038 0.0044 0.0044 0.0037 0.0046 0.0041 0.0041 0.0055

dev(H/D) 0.1290 0.0265 0.0590 0.1302 0.1900 0.0302 0.0449 0.0145 0.0194 0.0387 0.0326 0.0194 0.0236 0.0365 0.0178 0.0216 0.0399 0.0328 0.0348 0.0482

a u,,,,, values 0.25,0.7, 1.0, and 0.25 for model 3; 0.25, 0.25,0.2, and 1.0 for model 4; and 0.25, 0.7, and 1.1 for model 5. The order is that of the numbered bonds in the corresponding structures starting with C1 and proceeding clockwiie, skipping over bonds that are cut off in the models. Tunnel corrections calculated from the first term of the Bell equation are included in all isotope effects but change the effects by less than experimental error. 6 Isotope effects for very reactantlike and productlike species, respectively, in order to check that the force field leads to no isotope effect in the former and an equilibrium isotope effect in the latter case. All u,,,,, = 0.

effects gave poor fits to the secondary deuterium isotope effects, particularly at the 6-position. The basic difficultly in modeling the 6,6-D2 effect is that formation of a new c64,bond introduces new bending coordinates involving the 6-hydrogens. Other things being equal, this tends to depress the 6,6-D2 isotope effect below the value expected from rehybridization alone. This problem was dealt with by following a suggestion from Professor W.P. Huskey. The Huskey procedure involves using redundant sets of coordinates such that all bending coordinates of both the reactant and the product are represented in the transition structure. Thus, the C-H bonds at the Cq- and C6-positions are given outof-plane bending coordinates in the transition structure (even though these carbon atoms and the attached hydrogens do not lie in a plane) whose force constants vary with n(C0) and n(CC), disappearing at the reactant extreme in the former case and at the product extreme in the latter. That this procedure is physically reasonable is shown by the models in the third and second from the last lines in Table IV, which represent very reactantlike and productlike species, respectively. The former should give isotope effects near unity, which they do, and the latter should give values close to the equilibrium isotope effects. The 4,4-D2 effect (1.2874) is, in fact, close to the experimental estimate (1.27) of Gajewski and Conrad.' The 6,6-D2 effect (0.7244), however, is substantially below their estimate of 0.86. Nonetheless, our model is capable of matching the experimental kinetic isotope effects rather well. The only ways in which the agreement with the 6,6-D2 equilibrium isotope effect could be improved would be by using different algorithms to relate bond orders and force constants at the 4- and

6-positions or by reducing the C-C-H bending force constants at the 6-position to a value well below that which has given realistic results in previous contexts. No justification for either of these procedures is apparent, and they would certainly affect the fits to the kinetic isotope effects as well. The final results are shown in Table IV,which includes a large set of n(CO)/n(CC) values so as to make sure that all plausible combinations of bond orders, other than those in Table 111, were tried. The first five entries in Table IV are for completely synchronous models (the sum of n(C0) and n(CC) always equal to 1.0). They all give poor fits to the heavy-atom isotope effects, thoughonecase,n(CO) = 0.3,fits thesmndarydeuteriumisotope effects satisfactorily. The heavy-atom effects permit theexclusion of this and a number of other models which fit the deuterium isotope effects alone. The best models for all isotope effects are the ones with extents of bond breaking and bond making very similar to those for the aromatic Claisen rearrangement: n(CO)/ n(CC) of 0.5/0.2 and 0.4/0.2 for the aliphatic compared to OS/ 0.1 and 0.4/0.2 for the aromatic Claisen. Almost as good are the fits for 0.3/0.3 and 0.3/0.2. Poorer, but still not unreasonable, are the fits for 0.5/0.1 and 0.4/0.3. We can thus conclude that n(C0) lies in the range 0.34.5,while n(CC) lies in the range 0.14.3. The comparison of experiment and calculation thus indicates a transition structure which is reactantlike with respect to bond formation (10-30k) but central to productlikewith respect to bond cleavage (50-70%). The calculated equilibrium heavy-atom isotope effects (second from last line, Table IV) are all rather small, as expected from the fact that the total bond order to each of these atoms is the

Thermal Rearrangement of Allyl Vinyl Ether same in the product as in the reactant. Only the nature of the bonding changes, and there is, of course, no contribution due to reaction-coordinate motion. The irreversibility of the rearrangement precludes the direct experimental measurement of these effects. Indeed, the experimentalequilibriumisotope effects5 are estimates on the basis of results in related systems. Our model agrees very well with the conclusion of Gajewski and Conrad5 with respect to bond formation ('/& or 17%) but less well for bond cleavage (l/3, or 33%). It is evident, however, that all of our models with n(C0) in the range deduced by Gajewski and Conrad (0.60.7)fit the observed heavy-atom isotope effects rather poorly. Furthermore, it is difficult, qualitatively, to reconcile the large 3-180 and 4-14ceffcxts with a reactantlike 0 3 4 4 bond in the transition structure. The 4,4D2 effect alone allows a range of 0.7 to almost 0.3for n(CO), if an error of f0.03 in the experimental value is conceded. For these reasons, we believe that the extent of bond cleavage in the transition structure is 50% or more. In addition to giving information about bond orders in the transition structure, the heavy-atom isotope effects allow conclusions about the dynamics of the process. It is clear both from

J. Am. Chem. SOC.,Vol. 115, No. 14, 1993 5961 the magnitudes of the 3J80 and 4-14C effects relative to that of the 6-1- effect and from the greater couplings required to reproduce the former than the latter that cleavage of the 03-C4 bond contributes more to the reaction coordinate motion than formation of the C& bond. Finally, ab initio quantum mechanical calculations by Vance and co-workers6 find that in the transition structure, the C-O bond is stretched 37% from the length of that in 1 and that the forming C-C bond differs from the bond in 2 by 47%. As we have pointed out before? their bond distances translate approximately into bond orders of n(C0) = 0.18 and n(CC) = 0.09, even lower than those from our calculations. In summary, the BEBOVIB calculations arrive at a transition structure similar to but looser with respect to bond cleavage than that defined by Gajewski and Conrad. The heavy-atom KIE give, again, important information about the dynamics of the process.9Jo

Acknowledgments. We thank the National Science Foundation for support, Grants No. 89-19768and 88-18894.We also thank Professor W. P. Huskey for valuable suggestions.